Resumé

This paper compares and contrasts the Australian and Scandinavian welfare states with a view to demonstrating that, whilst the extent of welfare expenditures and the instruments of social policy vary quite markedly in these countries, policy outcomes in terms of levels of inequality and poverty and social protection are much more similar. In the course of this comparison, the paper casts serious doubts on the usefulness of both the prevailing paradigms for evaluating welfare state performance: measures of expenditure effort and measures of welfare decommodification. Instead, it is argued that welfare state performance can only be properly assessed, as Richard Tit muss pointed out many years ago, by evaluating the impact of fiscal and occupational welfare; in addition to the extent and character of the explicit expenditures of the state. When we broaden our conception of social policy in this way, Australia appears much less of a welfare state laggard than it is often taken to be and the oft mooted Scandinavian claim to welfare superiority is, perhaps, rather less compelling than is sometimes argued.

Introduction

This paper takes as its starting-point a contrast drawn from cross-national comparative analysis which is at the heart of much of the criticism directed at the character of the Australian welfare state. This contrast counterposes the Australian welfare state to those of Western Europe and, particularly, to those of the Scandinavian nations. On the basis of this contrast, two critiques of the Australian welfare state have been articulated, one resting on expenditure levels and the other on the institutional design of the welfare state. The arguments are that, compared with those of Scandinavia and many other European nations, the Australian welfare state is ungenerous and denies basic social rights of citizenship. The first contention tends to be heard from domestic critics of the Australian welfare state; the second has been a theme of recent comparative social policy scholarship emanating from Scandinavia.

My intentions here are twofold. First, I attempt to summarize the evidence and argument offered in favour of both contentions as a means of presenting the most commonly held views of Australia in the comparative welfare state literature. Second, I seek to demonstrate that both critiques suggest differences between the Australian and Scandinavian welfare

experience rather greater than those revealed by contrasts made in terms of welfare outcomes. My conclusion is that both critiques of the Australian welfare state exaggerate that country's exceptionalism because they rest on unduly limited conceptions of what constitutes social policy and of the means by which welfare objectives may be encompassed by the state.

The Expenditure Critique

The expenditure critique generally takes the form of a demonstration that, in comparison with most European nations, Australia's aggregate welfare expenditure has been extremely low throughout the post-war era. This is a case I have made in the context of a number of statistical analyses of the determinants of cross-national variation in social security expenditures (see Castles 1987 and 1992), but the location of the Australian contrast itself is one which scarcely requires much quantitative sophistication. Using Table 1 it is possible to compare Australian social security transfers expenditure with the Scandinavian and OECD averages.

For Australia, the story is one of relative decline from an already below average position. In 1960, according to this OECD data (OECD 1992, table 6.3), Australia was devoting fewer resources to social security than all of the richer OECD nations bar the United States, although it was spending somewhat more than a number of the very poorest nations. By 1990, whether other OECD countries were richer or poorer than Australia made no real difference: by that time, all of the OECD, bar Iceland, was devoting a greater percentage of national income to social security than was Australia. The frequently expressed view of Australia and Scandinavia (here counted as Denmark, Finland, Norway and Sweden) as polar opposites on the welfare state spectrum, is, however, only partially confirmed by this data. Whilst there is a clear difference in expenditure levels to be observed at both dates, and one which had increased very markedly over the period, it was not the case, either in 1960 or 1990, that the Scandinavian countries were in the vanguard of OECD transfers expenditure. In fact, the countries at the opposite extreme from Australia in this respect were,

in both periods, the core countries of Western Europe - France, Belgium,
the Netherlands, Luxembourg and Austria.

What goes for aggregate social security transfers also applies to its component elements. Although now somewhat dated, a comparison of the generosity of the major social insurance schemes of the welfare state in some fifteen OECD countries in 1960 and 1984 (see Castles 1987a) reveals Australia to have been spending less than the OECD mean in respect of all categories of welfare recipients at both dates. As was also true of the earlier contrast of aggregate expenditures, this more focused comparison of the generosity of particular schemes suggests a pattern in which, in 1960, Australia lagged well behind the OECD average in all respects and, to a somewhat lesser extent, behind the Scandinavian average in all respects bar pension level.

By 1984, the gap between Australia and both Scandinavia and the OECD in general had increased very markedly, with the Scandinavian countries now exceeding the OECD norm in all areas bar age and disability pensions. It is this relatively lower Scandinavian expenditure on age pensions compared to countries like the Netherlands, France and Austria which is the reason that the Scandinavian nations still do not feature as the OECD's biggest spenders in respect of transfers expenditure in the more recent period.

However, in those areas of welfare expenditure involving government final consumption, the case for Scandinavian welfare state leadership is quite unequivocal, whilst Australia's status as a laggard is far less pronounced than was the case in respect of social security transfers. Table 3 summarizes information for 15 OECD nations for which we have complete information on government final consumption expenditures on education, health, welfare and housing for 1990 (see OECD, 1992a).

As in the case of aggregate social security expenditure at the same date, Australian total final consumption expenditure on the welfare state in 1990 was appreciably less than the OECD average and was approximately half that of the Scandinavian nations, the shortfall showing up not so much in the areas of education and health as in the infrastructure of care, most particularly the salaries of welfare professionals and investment in and the running costs of residential facilities for the elderly and other categories of welfare clients. However, in this arena of expenditure, taking account of the way in which the OECD average is boosted by the Scandinavian countries' very high levels of spending, there is little real difference between the expenditure level of Australia and those of the non-Scandinavian countries of the OECD. Here the true exceptionalism is that of the very high levels of Scandinavian final consumption expenditure rather than the low level of Australian spending.

The proper conclusion from this very brief review of expenditure comparisons is that the Australia/Europe/Scandinavia comparison needs to be more nuanced than it often is. In respect of both aggregate transfers and the generosity of particular schemes, Australia lags behind both Europe and Scandinavia, but Australia's expenditures are less far removed from Scandinavian levels (especially at the earlier date) than from those of the core nations of Western Europe. On the other hand, in respect of final consumption expenditure, the big gap is between the Scandinavian vanguard and a less differentiated rearguard in which the West European nations do not conspicuously exceed Australian levels of expenditure.

The Institutional Design Critique

If welfare expenditure levels in Australia are substantially different from those prevalent in much of Europe, and particularly Scandinavia, it is no less true that Australia differs from these same countries in terms of the institutional design of its welfare state. Indeed, it has been a commonplace of authors commenting on the defining characteristics of the Australian social security system to single out institutional features which either singly

or together distinguish that system from those of most other Western nations. Often in criticism, although not invariably so (see Jones 1990), authors point to the unusual degree of selectivity of the Australian welfare benefits system (Australia is the only OECD country in which all social security transfers are means tested) or, more comprehensively, they note Australia's distinciiveness from other countries in providing flat-rate, subsistence level benefits from the general exchequer on a means-tested basis. All sorts of ills have been attributed to such an institutional structure of welfare provision. Flat-rate benefits cannot be sufficient to ensure a decent standard of life otherwise they will reduce work incentives. Selectivity through means tests leads to poverty traps and, it is sometimes argued, to a ghetto population or underclass of welfare dependents (the opposite argument from that generally advanced from the political Right, where it is the generosity of benefits which induces individuals to forsake gainful employment). Finance through general taxation gives the middle class a motive to resent the welfare state and selectivity reinforces that by ensuring that those who pay most benefit least.

Although the identification of Australian institutional distinctiveness clearly derives from an implicit comparison with the social policy systems of other nations, systematic comparison has, with only a very few exceptions (see Caim-Kaudle 1973), been impeded by a lack of standard data sets classifying cross-national variance in social policy instruments. In consequence, it has proved difficult to move beyond an approach in which diverse national experience is sorted into relatively undifferentiated ideal types. Now, however, a major data-gathering project on the characteristics of welfare state programmes in eighteen nations in the inter-war and postwar periods undertaken at the Stockholm Institute of Social Research, provides a potential basis for making far more explicit and detailed comparisons of the institutional characteristics of welfare systems and their implications for welfare provision.

The best-known account of this research, Gösta Esping-Andersen's The Three Worlds of Welfare Capitalism, explicitly rejects welfare expenditure as an appropriate basis for comparing national welfare levels on the ground that there is no guarantee that the input of government resources will be automatically translated into welfare outcomes. Instead of expenditure, his key concept is one of welfare "decommodification", by which is understood the extent to which the conferment of social entitlements leads to services being "rendered as a matter of right" with the consequence that "a person can maintain a livelihood without reliance on the market" (Esping-Andersen 1990, 22). Various features of the welfare system lead to greater or lesser decommodification, including benefit replacement rates, qualifications required for benefit, waiting days prior to benefit and the length of time a benefit is maintained. Most crucial of all, benefit level apart, is

the issue of coverage, since those excluded from benefit are seen as being
deprived of basic social rights of citizenship.

Given Australia's exceptional degree of social policy selectivity, and hence relatively low coverage ratios, and the low replacement rates implicit in subsistence level benefits, it is, perhaps, not surprising to find that Australia scores lowest of all countries on an overall decommodification index on which a grouping of three Scandinavian countries - Sweden, Norway, Denmark - score highest. Apart from his categorisation of welfare states and their individual programmes in terms of degrees of decommodification, Esping-Andersen also elaborates a series of indices measuring the extent to which social policy systems conform to conservative, liberal and socialist principles respectively. Conservatism describes the extent to which a system is characterized by corporatist and etatist features, liberalism is a measure of the degree of welfare privatization and of a focus on poor relief, whilst a socialist welfare system involves an emphasis on universalism and equality. In these contrasts, Australia again comes out as being very markedly different from Scandinavia in all respects other than conservatism, on which both score low and the core countries of Western Europe very high. According to Esping-Andersen (1990, 74), Australia possesses a highly liberal welfare system, whereas the degree of private welfare in Scandinavia is negligible (cf. Kuhnle & Selle 1992). On the other hand, the Scandinavian systems are seen to manifest the socialist virtues of universalism and equality to a very strong degree, whilst Australia, despite its equality of flat-rate benefits, and very much because of its high degree of selectivity, falls into an intermediate category on this index.

Many aspects of this institutional design contrast between the Australian and Scandinavian welfare states are well taken. Australian social policy has generally been imbued with a philosophy of targeting - which is just a modern way of describing and legitimizing selectivity and means testing - whilst Scandinavia, until recently, has been the region in which universal provision has been promoted most ardently.¹ Furthermore, Australian social policy has been concerned with the minimum required to allow individuals to lead a reasonable life, whereas the welfare states of Scandinavia have, at least in recent decades, emphasized income maintenance and high levels of income replacement rather than poverty alleviation (Heclo 1974). But whilst the institutional design contrast is real enough, its precise features and its implications for welfare outcomes are as much obscured as revealed by Esping-Andersen's various indices of welfare provision.

That is because, and to put it in terms of dramatic hyperbole, his analysis is rigged to come to the explicit conclusion that Australia is at the opposite extreme from the Scandinavian nations, whereas the difference, although quite real, is nothing like so extreme. The grounds for this assertion lie in

the footnotes and appendices to the tables in which Esping-Andersen
presents his indices. In the case of the decommodification index, we are
told that "in Australia, where (benefits) are based on a means test, we have

. . . . given the weight of 0.5 for population covered" (1990, 54) and, in the case of the universalism measure used to construct the socialism index, "that income-tested-based programs, such as the Australian and New Zealand unemployment and sickness benefit schemes, have been scored equal to O" (1990, 78). The argument for these operational procedures rests on the view that selective social policy diminishes social rights of citizenship by reducing entitlement and hence coverage. But the true measure of this shortfall from universal coverage must surely be the coverage actually achieved by social programmes, not some arbitrary, much lower, figure. The implication of Esping-Andersen's picture of Australian social policy is that no one has an entitlement to unemployment or sickness benefits, whereas, of course, almost the diametrically opposite is true. Because the sick and unemployed have no income, almost all of them pass the incomes test and qualify for benefit.

Two almost certain effects of removing this measurement bias from the Esping-Andersen indices would be that Australia would no longer be at the bottom of the decommodification scale and, more interesting still, that Australia would score as a strong socialist welfare state along with the Scandinavian nations. Arguably, it is the sheer improbability of this latter conclusion to a Scandinavian commentator, accustomed to assessing welfare virtue in terms of an ideology of universalism and seeing welfare development as a natural outcome of Social Democratic politics, that explains a weighting and distortion of the data unjustified by any other consideration.

As previously, in the case of expenditure differences, the evidence concerning differences in the institutional design of the Australian and Scandinavian welfare states suggests the possibility of a contrast far less black and white than usually drawn. Clearly, there are major differences in ideology concerning selectivity and universalism, and very substantial differences in social security benefit replacement ratios (which, although Esping-Andersen does not seem to realize it, is only to say that expenditure levels remain of major analytical importance, since replacement rates are, in effect, more or less equivalent to total expenditure divided by coverage). However, given that an unbiased account would lead us to the conclusion that, in some ways, the Australian social policy system is as "socialist" as those of Scandinavia, it would appear that there may well be other features of these social policy systems which are more similar. In what follows, I seek to identify some of these similarities. I do so, first, by questioning the prevailing conceptualization of welfare goals in the comparative literature and, second, by questioning that literature's very limited conceptualization of the nature of social policy.

The Welfare State and Equality

Esping-Andersen's point that expenditure inputs are not necessarily related to welfare outcomes is well taken. There is, however, a problem of at least equal magnitude in his alternative conceptualization of the goal of the welfare state as the decommodification of market relationships. Decommodification, despite its seemingly Marxist connotations, is a concept that owes more to English sociology (see Marshall 1975) than to socialist egalitarianism. Rather than seeing the aim of the welfare state in terms of social levelling and redistribution, it identifies the important goal as the creation of equality of status and the possession of identical social rights (including the equal right to the replacement of unequal incomes, as in the income replacement schemes of contemporary Scandinavia). Esping discussion is set in the context of what he describes as "a respecification of the welfare state" (1990, 21), and he is freely willing to concede that higher degrees of decommodification do not necessarily lead to greater redistribution (1990, 27), but, to the degree that this is so, he has clearly re-specified the concept in a quite different way from that in which it has commonly been used in much of the social policy literature. In that literature, the welfare state is about providing resources - income, education, health, etc. - to those who, for whatever reason, lack them and, given that such resources are intrinsically limited, it implies a welfare state which is fundamentally redistributive in character. If we view the institutional design of the welfare state in terms of features designed to achieve the goal of redistribution, the contrast between Australia, Europe and Scandinavia is transformed. Here, there is insufficient space for a complete account and I wish merely to make four illustrative points.

First, it is possible to measure the extent to which social security benefits are provided equally to all beneficiaries or, on the contrary, are related to earnings levels. Esping-Andersen himself produces such a measure of average benefit equality (1990, 70-71) and uses it as one element in his construction of the socialist welfare index. On this measure, the two most egalitarian nations are not Scandinavian, nor, indeed, European, but are, in fact, Australia and New Zealand. In this contrast, the Scandinavian countries are grouped relatively closely to the egalitarian Antipodes, with the Western European contributory social insurance systems markedly less equal and the United States most unequal of all. On Esping-Andersen's own account, the benefit equality indicator taps a vital aspect of welfare articulated according to socialist principles, and that this dimension is more pronounced in Australia than in Scandinavia, and much more so than elsewhere in the OECD, may be considered a real challenge to the implications of many of the other contrasts essayed in the comparative social policy literature.

Second, I wish very briefly to make some comments on the vexed question

of selectivity or means testing. To Esping-Andersen, it is this feature of the Australian welfare state which disqualifies any Australian claim to welfare state adequacy. But whilst such a case might possibly be arguable in terms of the weakness of social rights in Australia, it scarcely makes sense in terms of the direct impact of selectivity on resource redistribution. Income tests, quite contrary to the impact of earnings-related schemes, restrict benefit to those who are poorer. Their effect is necessarily redistributive in the sense that, for a given quantum of expenditure, the more focused the targeting to the least well off, the higher will be the degree of redistribution (see Åberg 1989). On that basis, a strong case can be made that Australia is amongst the most efficient social security systems of any in the OECD (see Mitchell, 1991). However, that case is not simultaneously a case that Australia's overall degree of redistribution is greater than all or most other Western nations. Most importantly, that is because the quantum of welfare expenditure is not, in fact, given and we know that Australia is quite close to the bottom of most social security expenditure distributions. The implications this has for Australia's position in the cross-national distribution of poverty and income inequality outcomes are examined below (see data in Table 4).

Another argument against the redistributive effects of selectivity, often pointed to by Scandinavian critics, including Esping-Andersen, suggests the possibility that there may be indirect negative effects on expenditure which outweigh the direct targeting impact, the supposed causal mechanism resting on diminished middle-class support for welfare spending from which only the less well off benefit. That argument may well have much to recommend it in principle, although, as far as I am aware, it has not been empirically demonstrated in any study, nor is it easy to see how it could be. It is, moreover, also relevant to mention that, of the countries which have a substantial means-tested component in their welfare systems, there are good reasons to suppose that the argument might apply less to the Australian and New Zealand cases than to the North American and British ones. Clearly, the deterrent effect on middle-class welfare support depends on the degree to which means tests exclude the middle class. Although one finds scant acknov/ledgement of the fact from the critics of selectivity, this varies very considerably from country to country. Australian means tests, far less than, say, American ones, exclude the middle class, which is why the coverage ratio of most schemes remains relatively high despite targeting. This point is also relevant to the social citizenship critique of Australian social policy. Where all but the very poor are excluded from benefit, it is quite reasonable to suggest that means tests abrogate basic social rights. That judgement may be less reasonable, where those excluded are primarily those who have the means to procure such rights through the market.

My third point is designed to illustrate the narrowness of the standard

conception of the bounds of social policy. In most of the literature, the institutions and instruments of social policy are more or less conterminous with social security transfers plus final consumption expenditure on health, welfare and community amenities. But as Titmuss long ago pointed out (Titmuss 1974), government expenditures are but one component of welfare effort. In addition, it is necessary to consider both fiscal welfare (how the state gives benefits by providing tax reductions for certain classes of beneficiaries and how it redistributes between income classes) and occupational welfare (the benefits accruing to some classes of employees in virtue of their employment).

At this juncture, I wish simply to point to the redistributive emphasis of the Australian tax system which makes it rather similar to the Scandinavian countries and rather different from much of the rest of Western Europe. As a colleague and I have demonstrated in recent research (Castles & Mitchell 1992), the cross-national variance of income and profits taxation as a percentage of GDP is very similar to the variance displayed by the benefit equality measure. Whilst the Scandinavian countries extract more of these progressive taxes than do Australia and New Zealand, and, in particular, use them far more systematically to claw-back some considerable share of universal income transfer benefits, both sets of nations are once again at an opposite extreme from much of Western Europe and the United States. Thus, to the extent that we widen our conception of social policy to include the redistributive impact of taxation, it is possible, once again, to identify a component of the Australian welfare state with a strong egalitarian thrust more comparable to Scandinavia than to the rest of the OECD.

My fourth and final point concerning the welfare state and equality is directed to the outcomes of the welfare state rather than the social security and taxation instruments by which it is achieved. On the basis of our argument so far, it would appear that conclusions as to the character of Australian welfare outcomes, insofar as they relate to poverty and income redistribution, are likely to be mixed. On the one hand, many features of the institutional design of Australia's welfare state - benefit equality, selectivity and progressive taxation - should be conducive to egalitarian outcomes. On the other, the evidence of all comparative studies of welfare expenditure suggests that Australia is at or near the bottom of the OECD distribution. The Australian system may be targeted, but the quantum of expenditure to be targeted is very low.

Assuming that expenditure and targeting considerations are of equal weight, we might expect that, in respect of poverty and inequality reduction, Australia would fall somewhere around the middle of an OECD distribution, in which we would expect the Scandinavian countries to have the highest values in virtue of high expenditure levels combined with an

institutional design of welfare which also strongly emphasizes egalitarian objectives. In such a comparison, we might expect the Western European nations with high expenditure levels, but with social policy instruments with only a limited redistributive emphasis, to manifest outcomes not all that dissimilar from those of Australia. Finally, we might predict that the North American countries and Switzerland, which spend relatively little and do not make any great redistributive effort, would emerge as the OECD laggards.

Data calculated from the Luxembourg Incomes Study by Mitchell (1991), and presented in Table 4, broadly confirm these expectations, irrespective of whether the comparisons relate to the number of families in poverty, the size of the poverty gap or the net redistributive effect of the social security and taxation systems. In these comparisons, the Scandinavian countries invariably cluster towards the top of the distributions and the North American countries and Switzerland towards the bottom. Western Europe and Australia come in between, with the only departure from our predicted ranking being something of a tendency (pronounced in the case of poverty rates) for Australia to fall rather behind the Western European nations. This latter may well suggest that, ultimately, high expenditure is a somewhat more important determinant of redistributive outcomes than is an egalitarian institutional design of the social policy system. Whether or not that is the case, these comparisons of redistributive outcomes do clearly demonstrate that the standard picture of Australia as the polar opposite of Scandinavia is quite misconceived, and the view of Australian welfare standards as markedly inferior to those of some, supposedly,

archetypal "European welfare state" no less so. Australia is different in many ways from some or all of these countries in its distinctive emphasis on low but highly targeted expenditures and taxes, but it is far less distinctive in the final consequences of this mix - levels of poverty and inequality which are generally only just below OECD means.

Social Policy by Other Means

In what I have said so far, I have argued that a focus on redistributive outcomes quite strongly modifies the conclusion that Australia is a world of welfare apart from Europe and Scandinavia. But there is another, perhaps more fundamental, reason why an analysis directed to welfare outcomes is likely to modify conclusions that Australia is in a separate league from other welfare states. A focus on outcomes always suggests the possibility that a diversity of policy means disguises a basic similarity of policy goals and outcomes. That case has already been made several times in the earlier discussion, with tight targeting of a small quantum of expenditure being argued to be roughly equivalent to weak targeting of a much larger quantum of expenditure, and a progressive taxation system being regarded as functionally equivalent to a selective social security system.

These instances are far from isolated ones. In concluding, I wish briefly to examine three further examples of the way in which welfare goals, which in the conventional account are scarcely catered for by the Australian system, are achieved via routes quite different from those in most other Western countries.² An emphasis on the diversity of routes to similar policy goals is merely an academic way of restating the old English homily about there being many ways to skin a cat. My point is that Australians skin the welfare cat rather differently from most other nations. Showing how will, I hope, illustrate the character, the developmental trajectory and the real nature of the distinctiveness of the Australian welfare state.

Let me start with the example of sickness provision. On the surface, it is the familiar story of typically ungenerous Australian provision. In 1984, Australia spent 0.18 of one percent of GDP on social insurance sickness benefits and Australian workers averaged three days a year absence taken on such benefits. In contrast, the average expenditure for the OECD was 0.78 percent of a percent of GDP and the average number of days absence was 11. In Sweden, the unquestioned big spender in this policy area, 2.83 percent of GDP was spent and 21 days of absence taken. For Scandinavia as a whole, average expenditure was 1.61 percent of GDP and the average number of sick leave days taken was 12 (all data in this paragraph from Kängas 1991).

So the question is: is Australia really so ungenerous in its sickness provision? And the answer is, as any Australian will tell you, quite obviously, no. Australia has what it calls "sickies" and Australian employers will tell you that Australian workers "chuck" them with gay abandon and little observation of the niceties of medical certificates. In faet, sickies are not necessarily the corrupt practice employers frequently make them out to be, but they do take Australia some way up the sickness provision league table, costing perhaps around 0.8 of a percent of GDP and adding an average of three further days per worker to annual absence figures (see Castles 1992). On these figures, Australia becomes far more like the average OECD country in terms of sickness provision.

We can now let the way we skin the Australian welfare cat out of the bag. Other countries provide sickness provision through social insurance, but for the most part Australia does not. Rather it is provided by employer coverage mandated in wage awards made by the industrial tribunals which have existed in various forms, at both State and Federal level, since very early on this century. The provision is, in other words, via Titmuss's route of occupational welfare (Titmuss 1974). To overseas observers unfamiliar with Australian institutions, these arrangements may look less secure than the rights conferred by social insurance benefits, but since provision is equally guaranteed by an instrumentality of the state, it should properly be regarded as being just as much part of the welfare state as any other of the outputs of government.

Moreover, once wage awards are regarded as a potential source of state welfare, we can start dating the Australian welfare state rather earlier than do most European commentators. The first Commonwealth award containing compulsory provision of sick days was the Engineers' case of 1920 and this provision soon spread to other Commonwealth and State awards. Since compulsory sickness provision in Europe was rare prior to World War 11, Australia suddenly becomes an international welfare pioneer rather than a welfare laggard in this perspective.

A similar story can be told in the case of my second example about a much more recent innovation: namely awards-based superannuation and its subsequent extension by legislation. The similarity is not that Australia is a pioneer, but rather that we encounter provision through the mechanism of wage-setting and the industrial relations system and that, once again, the preferred Australian strategy of the Australian welfare state is via occupational welfore rather than governmental provision. Here, too, is a scheme mandated by the state which provides a welfare benefit - compulsory employer-funded earnings-related superannuation - which elsewhere in much of the rest of the OECD is a part of the social security system. We may also note a feature of this type of provision, which makes it particularly appealing to governments confronted, as was the Hawke

Labour government in Australia, by the perceived need to contain public expenditure: the fact that superannuation paid by employers does not feature in National Accounts either as an expenditure or a taxation item. In other words, governments can gain the credit for increasing welfare without the perceived odium of spending and taxing more.

The initial 3 percent level of the wage bill at which awards-based superannuation contributions were set was tiny compared with European/ Scandinavian earnings-related superannuation/pensions schemes established for many decades. However, the eventual target of 12 percent approximates to roughly 8 percent of GDP. If Australian and OECD data sources do not include these amounts as part of Australia's welfare effort they will seriously misrepresent that effort in comparative perspective.³

The issue of superannuation is particularly interesting because the advent of earnings-related schemes may hold the key to a process of convergence between the Australian and Scandinavian age-pensions systems. That is not merely because award superannuation in Australia will lead to the wider availability of high replacement age-pensions along lines mapped out in Scandinavia over several decades, but also because processes set in motion by the adoption of superannuation schemes in the Scandinavian nations now seem likely to erode welfare universalism in those countries, leading to systems only marginally less selective than the Australian one. The Scandinavian age-pension systems consist fundamentally of three elements: a basic and universal pension, a superannuation benefit based on past earnings and a supplementary pension paid only where the level of superannuation is inadequate (i.e. it is means tested against superannuation). As recent research has noted (Øverbye 1992), the long-term trend in all the Scandinavian countries has been for the universal basic component to decline and the supplementary component to increase, the logic being precisely that informing Australian means testing (i.e. a reluctance to pay large sums of public money to those already in receipt of adequate retirement income). In Finland, according to this research, the basic amount is now down to 8 percent of average industrial wages (calculated from Øverbye 1992) (Tables 1 and 2), already a somewhat slim margin on which to assert a major contrast between that Scandinavian country's supposedly universal welfare state and Australia's selective one.

My third example takes us back to the very origins of the distinctiveness of the Australian welfare state: the creation through Mr Justice Higgins' Harvester Judgement of 1907 of a wage award minimum designed to achieve social policy goals or, in Higgins' words, that he could not think of "any other standard appropriate" to assess "a fair and reasonable wage"

. . . "than the normal needs of the average employee regarded as a human being living in a civilised community". (Clark 1981, 286). To Higgins, that standard was seven shillings a day, an amount supposed to be sufficient to support a wife and two or three children in frugal comfort.

As I have argued at some length elsewhere, this meant that as early as the first decade of this century, Australia, unlike most other modern nations at the time, had a real welfare state, albeit one that was rather different from the social insurance states that came to typify European development. The Australian welfare state was what I have called "a wage earners welfare state" in offering social protection particularly focused on those who participated in the workforce and providing only far more vestigial, somewhat ungenerous and means-tested support for those outside. That was the pattern of provision at the beginning of the century and, in many ways, it remains the pattern of provision today, as the previous examples of sickness provision and superannuation clearly demonstrate. In that sense, the Australian welfare state was quite distinctive in its form and trajectory, even if its outcomes for much of the period were not, seemingly, in any way inferior to those of the European and Scandinavian welfare states.

Indeed, in terms of outcomes, it is possible to argue, that much of Australia's reputation as an egalitarian society in the first half of this century was built around the notion of the "wage-earners' welfare state" and its associated public policy strategies (see Castles 1985 and 1988). It is, however, also arguable that in the last quarter century, as major groups of welfare clientele outside the employment structure have emerged - conspicuously, the unemployed, single parents and the children of both groups - this strategy of social protection has become progressively outmoded, leading to appreciably higher levels of poverty and inequality than in Scandinavia and to an egalitarian performance rather below the OECD mean.

The fundamental characteristic of the "wage-earners' welfare state" is the provision of social protection by means of occupational welfare mandated by the state. Such a strategy may be particularly pronounced in Australia, but it is not unique to that country (see Castles & Mitchell 1992), and there are some signs that it is becoming the preferred strategy of welfare containment under the circumstances of financial restraint faced by many countries in the 19905. That strategy is quite different from one depending on government expenditure and its importance is generally obscured where the focus of research is on the quantum of such expenditure and/or the policy instruments by which that expenditure is delivered. Irrespective of whether Australia's welfare state is a relic from an earlier age or a portent of the shape of things to come, what is certain is that the only way we can sensibly compare and contrast its nature vis-å-vis other systems is by focusing on the extent to which alternative policy routes may lead to similar, although by no means identical, policy outcomes.

---

NOTES

1. Obviously, there have been differences between the countries, Marklund (1988) arguing that Sweden and Finland have more universalistic schemes than Denmark and Norway. In the disastrous employment situation of the present time, there is active consideration of income testing of at least some benefits in all these countries, not least Sweden. Indeed, judging by parliamentary debates in Sweden in the early part of 1993, the only root and branch political supporter of the universalist principle is the Liberal Party.

---

1. Obviously, there have been differences between the countries, Marklund (1988) arguing that Sweden and Finland have more universalistic schemes than Denmark and Norway. In the disastrous employment situation of the present time, there is active consideration of income testing of at least some benefits in all these countries, not least Sweden. Indeed, judging by parliamentary debates in Sweden in the early part of 1993, the only root and branch political supporter of the universalist principle is the Liberal Party.

2. As I have argued elsewhere (Castles 1985), the New Zealand welfare state shares many features in common with that of Australia.

3. Quite extraordinarily, the Finnish superannuation system has many of the same characteristics as those now emerging in Australia. The OECD does not count Finland's very considerable superannuation expenditure as a part of social security transfers, but Finnish social policy experts most certainly do.

REFERENCES

Caim-Kaudle, P. R. 1973. Comparative Social Policy and Social Security. London: Martin
Robertson.

Castles, F. G. 1985. The Working Class and Welfare. Sydney: Allen & Unwin.

Castles, F. G. 1987. 'Thirty Wasted Years: Australian Social Security Development, 1950-80,
in Comparative Perspective', Politics 67-74.

Castles, F. G. 1987a. 'Trapped in an Historical Cul-de-sac: The Prospects of Welfare Reform in Australia', in Social Welfare in the Late 1980s: Reform, Progress or Retreat? University of New South Wales, Social Welfare Research Centre: Reports and Proceedings, No. 65, 91-101.

Castles, F. G. 1988. Australian Public Policy and Economic Vulnerability. Sydney: Allen &
Unwin.

Castles, F. G. 1992. 'On Sickness Days and Social Policy', Australian & New Zealand Journal
of Sociology 28, 29-44.

Castles, F. G. and Mitchell, D. 1992. 'Identifying Welfare State Regimes: The Links between
Politics, Instruments and Outcomes'. Governance 5 (1), 1-26.

Clark, M. 1981. A History of Australia, Vol. 5. Melbourne: Melbourne University.

Heclo, H. 1974. Modern Social Politics in Britain and Sweden. New Haven: Yale University
Press.

Jones, M. A. 1990. The Australian Welfare State. 3rd edn. Sydney: Allen & Unwin.

Kängas, O. 1991. The Politics of Social Rights. Stockholm: Swedish Institute of Social
Research.

Kuhnle, S. & Selle P. 1992. Government and Voluntary Organizations. Aldershot: Avebury.

Marklund, S. 1988. Paradise Lost? The Nordic Welfare States and the Recession 1975-1985.
Lund: Arkiv.

Marshall, T. H. 1975. Social Policy, rev. ed. London: Hutchinson.

OECD, 1992a. OECD in Figures, Supplement to the OECD Observer, No. 176, Paris.

Øverbye, E. 1992. Public or Private Pensions?: Pensions and Pension Politics in the Nordic
Countries. Oslo: Institutt for sosialforskning, No. 6.

Åberg, R. 1989. 'Distributive Mechanisms of the Welfare State - a Formal Analysis and an
Empirical Application', European Sociological Review, 5, 167-82.